Finland AI Regulatory Sandbox Proposal
Government proposes establishing AI regulatory sandbox
Finland
RAI-FI-NA-REGULAT-2026Finland AI Regulatory Sandbox Proposal is Proposed in Finland, according to eur-lex.europa.eu. We have not yet been able to confirm the status.
BillGovernance and OversightSafety, Testing, and EvaluationRisk ManagementFinland proposes a Traficom-coordinated AI regulatory sandbox for supervised testing under legislation complementing the EU AI Act.
Summary
Finland’s Government proposes national legislation establishing an AI regulatory sandbox to complement the EU AI Act. Businesses and other AI developers would test solutions under official supervision, with Traficom responsible for establishing, maintaining, developing, and coordinating the environment.
Full article
Read full text ↗Overview
Finland’s Government proposes establishing a national AI regulatory sandbox through legislation complementing the EU AI Act. The proposed environment would allow businesses and other AI system developers to try new solutions under the supervision of authorities. The Ministry of Economic Affairs and Employment published the announcement, Government proposes establishing AI regulatory sandbox, on 22 October 2026. The initiative combines support for AI development with an official setting for examining regulatory questions and risks associated with rapidly advancing technology.
The proposed legislative initiative concerns the creation and organisation of a supervised testing environment. Its central institutional feature is the assignment of responsibility to the Finnish Transport and Communications Agency Traficom. Traficom would establish, maintain, develop, and coordinate the national sandbox. The announcement’s English heading identifies the initiative, while the proposed national legislation would provide the legal framework for its establishment. The initiative is a proposed bill, and its objectives centre on supervised experimentation, trustworthy and safe AI, and assistance with identifying applicable obligations.
Purpose and Policy Objectives
The sandbox would give businesses and other AI system developers an environment in which to try new solutions with authorities involved in supervision. This brings innovation and regulatory engagement into the same setting. Developers could explore the operation of an AI solution while considering the regulatory questions arising from its development and testing. The proposal responds to the rapid advancement of AI by combining an opportunity for experimentation with attention to associated risks. Its purpose is therefore both developmental and regulatory: supporting new solutions while promoting responsible consideration of their effects.
Traficom describes the objectives of the AI regulatory sandbox in terms of promoting trustworthy and safe AI and helping organisations identify their obligations under the AI framework. These objectives connect technical development with compliance learning. Trustworthiness and safety guide the environment’s intended direction, while assistance with obligations helps organisations understand the legal setting for their activities. Together, these aims position the sandbox as a supervised environment for learning about an AI system and its regulatory context, rather than merely a place for demonstrating technical performance.
Participants and Development Activities
The proposed environment is intended for businesses and other AI system developers. That description encompasses commercial development alongside development by other types of organisations. The common feature is work on AI solutions that could be tried in a supervised setting. The proposal’s focus on developers places experimentation close to the organisations designing and building systems. It also connects regulatory engagement to development activities, allowing questions about a solution’s intended operation, potential risks, and applicable obligations to be considered alongside its technical characteristics.
For voluntary preparation, a prospective participant can describe the AI solution it intends to test, the purpose of the experiment, and the questions it wants to discuss with authorities. A useful project description can distinguish the system’s intended function from the particular activity proposed for testing. It can also identify the people or organisations involved in development, the anticipated setting for the experiment, and the expected learning outcomes. These are practical planning steps for engaging with a supervised testing environment, rather than a statement of statutory application requirements.
Governance and Institutional Responsibilities
The Government proposes that Traficom be responsible for establishing, maintaining, developing, and coordinating the national AI regulatory sandbox. Establishment concerns creating the environment. Maintenance concerns its continuing operation, while development concerns its improvement over time. Coordination concerns organising the sandbox’s activities. Taken together, these responsibilities give Traficom a central administrative role in the proposed national arrangement. The combination also connects the initial creation of the sandbox with its longer-term operation and evolution as AI technologies and regulatory questions develop.
The Ministry of Economic Affairs and Employment is the issuing authority for the government announcement. Traficom is the proposed implementing authority for the sandbox itself. These roles distinguish the announcement of the legislative initiative from the responsibilities envisaged for operating the environment. Official supervision is a defining feature of the proposal, and coordination would organise the activities associated with that supervision. For organisations planning engagement, the institutional structure makes Traficom the central agency identified for the establishment and management of the proposed sandbox.
Supervised Testing and Evaluation
The proposed sandbox would enable businesses and other developers to try new AI solutions under official supervision. Testing and evaluation are therefore central focus areas. The environment would connect experimentation with the involvement of authorities, bringing regulatory considerations into the development process. Its practical purpose is to provide a setting for trying solutions and learning from their operation. The emphasis on supervision distinguishes the proposed sandbox from an entirely private development exercise and connects experimentation with public oversight.
As voluntary project planning, an organisation can formulate clear testing objectives, identify the behaviour it wants to examine, and describe how it intends to observe the results. It can separate technical questions from questions requiring regulatory guidance, while recognising that the two may interact. For example, the intended use of a system may shape both the relevant experiment and the obligations the developer needs to understand. A structured description of objectives and observations can support useful discussions with authorities. Such preparation serves the sandbox’s testing purpose without turning illustrative planning practices into legal duties.
Risk Management and Safe AI
Risk management is a central topic because the initiative combines experimentation with attention to risks associated with rapidly advancing AI. Traficom’s stated objective of promoting trustworthy and safe systems reinforces this emphasis. The proposed environment would allow development and regulatory engagement to occur together, supporting consideration of risks as solutions are tried. Safety and trustworthiness are policy objectives for the sandbox, while official supervision provides the institutional setting in which developers and authorities would engage with experimentation.
For voluntary preparation, developers can identify foreseeable risks associated with their proposed experiment and distinguish risks arising from the system from risks arising from the testing setting. A project description can explain the intended users, the circumstances in which the system would operate, and the observations that would help assess its behaviour. Developers can also record questions about safeguards for discussion with authorities. These practical steps connect the proposed sandbox’s safety objective with a concrete development project, while remaining planning recommendations rather than prescribed assessment methods or mandatory risk thresholds.
Relationship to the EU AI Act
The Government proposes national legislation that would complement the EU AI Act. This is the initiative’s principal relationship to another legal instrument and its central point of international alignment. The relevant EU instrument is Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence. The Finnish proposal would establish a nationally organised sandbox within that broader EU regulatory context. Traficom’s proposed responsibilities concern the Finnish environment, while the legislative approach connects its establishment to the EU AI framework.
The national and EU dimensions address different aspects of the initiative. National legislation would organise the Finnish sandbox, and the EU AI Act supplies the broader artificial intelligence regulatory setting that the proposal would complement. Traficom’s objective of helping organisations identify their obligations connects these dimensions at the level of participating developers. For voluntary legal planning, an organisation can consider its proposed testing activities alongside applicable EU and national requirements. The appropriate planning approach is to treat experimentation and regulatory understanding as related tasks throughout development.
Operational Preparation and Compliance Learning
The proposed implementation approach combines national legislation with a designated agency responsible for creating and operating the environment. Traficom’s roles cover establishment, continuing maintenance, development, and coordination. These functions give the proposal an organisational structure extending beyond the initial creation of a testing setting. The compliance-learning objective adds a practical purpose to that structure: helping organisations identify obligations as they work on AI solutions. The environment would therefore bring together the system being developed, the planned experiment, and the regulatory questions relevant to that work.
Voluntary preparation can include a concise description of the system, testing objectives, anticipated risks, and questions requiring supervisory guidance. Organisations can also organise records of their intended experiment and the observations they expect to make. Before undertaking a project through the sandbox, a practical planning step is to consult official participation arrangements and obtain the relevant acceptance for the proposed activity. Legal planning can address the effect of participation, responsibility for the experiment, and applicable requirements alongside technical preparation. These actions are recommendations for prospective participants, not statutory duties attributed to the proposal.
Implementation Timeline
| Milestone | Date | Significance |
|---|---|---|
| Government announcement published | 2026-10-22 | The Ministry of Economic Affairs and Employment announced the proposal to establish an AI regulatory sandbox through national legislation complementing the EU AI Act. The announcement was published at 14.28. |
| Proposed establishment and operation | Proposed legislative implementation | Traficom would establish, maintain, develop, and coordinate the national testing environment. |
Planning Checklist
| Planning Area | Voluntary Preparatory Action |
|---|---|
| Legal instrument and participation | Consult official legislation and participation arrangements when planning an application or testing activity. |
| Authority and coordination | Use Traficom’s official information to plan engagement with the agency identified to establish and coordinate the environment. |
| Project description | Describe the AI solution, its intended function, the planned experiment, and the development questions to be examined. |
| Testing objectives | Identify the behaviour to be observed, the expected learning outcomes, and the questions requiring supervisory guidance. |
| Risks and safeguards | Consider foreseeable risks associated with the system and testing setting, and prepare questions about suitable safeguards. |
| Documentation and learning | Organise planned observations, testing aims, and questions about applicable obligations to support discussions with authorities. |
| Legal effects and remedies | Review applicable participation terms, responsibility for testing activities, liability arrangements, enforcement provisions, and available remedies as part of legal planning. |
| Other applicable requirements | Consider relevant EU and national obligations, including requirements connected with data protection and sector-specific activities. |
| Cross-border projects | Review territorial participation arrangements and the treatment of testing outcomes when planning activities involving more than one jurisdiction. |
Sources and References
| Source | Type |
|---|---|
| Government proposes establishing AI regulatory sandbox | official |
| AI regulatory sandbox | official |
| Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence | official |
More AI regulation in Finland
© Regulations.AI — created on 21 Nov 2026 using Gemini 3.6 Flash